
import pandas as pd
import numpy as np
import plotly.express as px
import seaborn as sns
import matplotlib.pyplot as plt
from collections import Counter
from collections import defaultdict
import itertools
from wordcloud import WordCloud
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from textblob import TextBlob


df = pd.read_csv("North_America-United_States.csv")


df.head()


df.shape

(42253, 31)


df.nunique()

data_id             42253
iso                     1
event_id_cnty       42253
event_id_no_cnty    42253
event_date            905
year                    3
time_precision          3
event_type              2
sub_event_type          5
actor1                  6
assoc_actor_1        5428
inter1                  2
actor2                 26
assoc_actor_2         310
inter2                  8
interaction            14
region                  1
country                 1
admin1                 51
admin2               1224
admin3                  0
location             4207
latitude             5134
longitude            5195
geo_precision           3
source              14344
source_scale           20
notes               42250
fatalities              4
timestamp             374
iso3                    1
dtype: int64


df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 42253 entries, 0 to 42252
Data columns (total 31 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   data_id           42253 non-null  int64  
 1   iso               42253 non-null  int64  
 2   event_id_cnty     42253 non-null  object 
 3   event_id_no_cnty  42253 non-null  int64  
 4   event_date        42253 non-null  object 
 5   year              42253 non-null  int64  
 6   time_precision    42253 non-null  int64  
 7   event_type        42253 non-null  object 
 8   sub_event_type    42253 non-null  object 
 9   actor1            42253 non-null  object 
 10  assoc_actor_1     32392 non-null  object 
 11  inter1            42253 non-null  int64  
 12  actor2            3882 non-null   object 
 13  assoc_actor_2     1152 non-null   object 
 14  inter2            42253 non-null  int64  
 15  interaction       42253 non-null  int64  
 16  region            42253 non-null  object 
 17  country           42253 non-null  object 
 18  admin1            42253 non-null  object 
 19  admin2            42186 non-null  object 
 20  admin3            0 non-null      float64
 21  location          42253 non-null  object 
 22  latitude          42253 non-null  float64
 23  longitude         42253 non-null  float64
 24  geo_precision     42253 non-null  int64  
 25  source            42253 non-null  object 
 26  source_scale      42253 non-null  object 
 27  notes             42253 non-null  object 
 28  fatalities        42253 non-null  int64  
 29  timestamp         42253 non-null  int64  
 30  iso3              42253 non-null  object 
dtypes: float64(3), int64(11), object(17)
memory usage: 10.0+ MB


df.isnull().sum()

data_id                 0
iso                     0
event_id_cnty           0
event_id_no_cnty        0
event_date              0
year                    0
time_precision          0
event_type              0
sub_event_type          0
actor1                  0
assoc_actor_1        9861
inter1                  0
actor2              38371
assoc_actor_2       41101
inter2                  0
interaction             0
region                  0
country                 0
admin1                  0
admin2                 67
admin3              42253
location                0
latitude                0
longitude               0
geo_precision           0
source                  0
source_scale            0
notes                   0
fatalities              0
timestamp               0
iso3                    0
dtype: int64


df.drop(columns=["admin3"], inplace=True)


df["event_date"] = pd.to_datetime(df["event_date"])


df.head()


df['year'] = df['event_date'].dt.year.astype(str)
yearly_counts = df['year'].value_counts().reset_index()
yearly_counts.columns = ['year', 'count']

fig = px.pie(yearly_counts, names="year", values="count", title="Number of Protests per Year")
fig.show()


df['year_month'] = df['event_date'].dt.strftime('%Y-%m')

monthly_protest_counts = df.groupby('year_month').size().reset_index(name='protest_count')


fig = px.bar(monthly_protest_counts, x='year_month', y='protest_count',
             title='Monthly Protest Count Per Year',
             labels={'year_month': 'Year-Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()


event_counts = df['event_type'].value_counts().reset_index()
event_counts.columns = ['event_type', 'count']

fig = px.pie(event_counts, names="event_type", values="count", title="Count of Event types")
fig.show()


sub_event_counts = df['sub_event_type'].value_counts().reset_index()
sub_event_counts.columns = ['sub_event_type', 'count']

fig = px.pie(sub_event_counts, names="sub_event_type", values="count",hole = 0.3, title="Count of Sub Event types")
fig.update_xaxes(tickangle=-90)

fig.show()


actor1_counts = df['actor1'].value_counts().reset_index()
actor1_counts.columns = ['actor1', 'count']

fig = px.bar(actor1_counts, x="actor1", y="count", title="Count of Actor1")
fig.update_xaxes(tickangle=-45)
fig.show()


assoc_actors = df['assoc_actor_1'].dropna().tolist()

assoc_actors_cleaned = []
for actor in assoc_actors:
    if ';' in actor:
        assoc_actors_cleaned.extend(actor.split(';'))
    else:
        assoc_actors_cleaned.append(actor)

assoc_actors_cleaned = [actor.strip() for actor in assoc_actors_cleaned]

assoc_actor_counts = Counter(assoc_actors_cleaned)

sorted_actors = sorted(assoc_actor_counts, key=assoc_actor_counts.get, reverse=True)

top_50_actors = sorted_actors[:50]
top_50_df = pd.DataFrame.from_dict({actor: assoc_actor_counts[actor] for actor in top_50_actors}, orient='index', columns=['Count']).reset_index()
top_50_df.rename(columns={'index': 'Actor'}, inplace=True)
#fig.update_xaxes(tickangle=45)

fig = px.bar(top_50_df, x='Actor', y='Count', title='Top 50 Associated Actors')
fig.show()


actor2_counts = df['actor2'].value_counts().reset_index()
actor2_counts.columns = ['actor2', 'count']

fig = px.bar(actor2_counts, x="actor2", y="count", title="Count of Actor2")
fig.update_xaxes(tickangle=-45)
fig.show()


assoc_actors_2 = df['assoc_actor_2'].dropna().tolist()

assoc_actors_2_cleaned = []
for actor in assoc_actors_2:
    if ';' in actor:
        assoc_actors_2_cleaned.extend(actor.split(';'))
    else:
        assoc_actors_2_cleaned.append(actor)

assoc_actors_2_cleaned = [actor.strip() for actor in assoc_actors_2_cleaned]

assoc_actor2_counts = Counter(assoc_actors_2_cleaned)

sorted_actors_2 = sorted(assoc_actor2_counts, key=assoc_actor2_counts.get, reverse=True)

top_50_actors = sorted_actors_2[:50]
top_50_df = pd.DataFrame.from_dict({actor: assoc_actor2_counts[actor] for actor in top_50_actors}, orient='index', columns=['Count']).reset_index()
top_50_df.rename(columns={'index': 'Actor'}, inplace=True)

fig = px.bar(top_50_df, x='Actor', y='Count', title='Top 50 Associated Actors 2')
fig.update_xaxes(tickangle=-45)
fig.show()


interaction_counts = df['interaction'].value_counts().reset_index()
interaction_counts.columns = ['interaction', 'count']

fig = px.pie(interaction_counts, names="interaction", values="count",hole = 0.3, title="Count of Protests per interaction")
fig.update_xaxes(tickangle=-90)
fig.show()


admin1_counts = df['admin1'].value_counts().reset_index()
admin1_counts.columns = ['admin1', 'count']

fig = px.bar(admin1_counts, x="admin1", y="count", title="Count of Admin1")
fig.update_xaxes(tickangle=-45)
fig.show()


admin2_counts = df['admin2'].value_counts().reset_index()
admin2_counts.columns = ['admin2', 'count']

top50_admin2_counts = admin2_counts.head(50)

fig = px.bar(top50_admin2_counts, x="admin2", y="count", title="Top 50 Count of Admin2")
fig.update_xaxes(tickangle=-45)
fig.show()


location_counts = df['location'].value_counts().reset_index()
location_counts.columns = ['location', 'count']

top50_location_counts = location_counts.head(50)

fig = px.bar(top50_location_counts, x="location", y="count", title="Top 50 Count of location")
fig.update_xaxes(tickangle=-45)
fig.show()


sources = df['source'].dropna().tolist()

sources_cleaned = []
for actor in sources:
    if ';' in actor:
        sources_cleaned.extend(actor.split(';'))
    else:
        sources_cleaned.append(actor)

sources_cleaned = [actor.strip() for actor in sources_cleaned]

sources_counts = Counter(sources_cleaned)

sorted_sources = sorted(sources_counts, key=sources_counts.get, reverse=True)

top_50_actors = sorted_sources[:50]
top_50_df = pd.DataFrame.from_dict({actor: sources_counts[actor] for actor in top_50_actors}, orient='index', columns=['Count']).reset_index()
top_50_df.rename(columns={'index': 'Sources'}, inplace=True)

fig = px.bar(top_50_df, x='Sources', y='Count', title='Top 50 Associated Sources')
fig.update_xaxes(tickangle=-45)
fig.show()


source_scale_counts = df['source_scale'].value_counts().reset_index()
source_scale_counts.columns = ['source_scale', 'count']

top50_source_scale_counts = source_scale_counts.head(50)

fig = px.bar(top50_source_scale_counts, x="source_scale", y="count", title="Top 50 Count of source_scale")
fig.update_xaxes(tickangle=-45)
fig.show()


df = df[df['actor1'].str.contains('United States', case=False)]
df.shape

(42238, 31)


df.describe()


pair_freq = defaultdict(int)

for _, row in df.dropna(subset=['assoc_actor_1', 'assoc_actor_2']).iterrows():

    actors_1 = [actor.strip() for actor in row['assoc_actor_1'].split(';')]
    actors_2 = [actor.strip() for actor in row['assoc_actor_2'].split(';')]

    for actor_1, actor_2 in itertools.product(actors_1, actors_2):
        pair_freq[(actor_1, actor_2)] += 1

pair_freq_df = pd.DataFrame(list(pair_freq.items()), columns=['Actor Pair', 'Frequency'])

pair_freq_df_sorted = pair_freq_df.sort_values(by='Frequency', ascending=False).reset_index(drop=True)

pair_freq_df_sorted.head(10)


pair_freq_df_sorted['Actor Pair Str'] = pair_freq_df_sorted['Actor Pair'].apply(lambda x: f"{x[0]} & {x[1]}")
# As the values are stored in pairs, using the lamda function we are splitting
# them and spearating the Pairs using an & for understanding.

plt.figure(figsize=(10, 8))
sns.barplot(x='Frequency', y='Actor Pair Str', data=pair_freq_df_sorted.head(20))
plt.title('Top 20 Most Frequent Actor Pairs')
plt.xlabel('Frequency')
plt.ylabel('Actor Pairs')
plt.tight_layout()
plt.show()


df['interaction'] = df['interaction'].astype(str)
interaction_counts = df['interaction'].value_counts().reset_index()
interaction_counts.columns = ['interaction', 'count']
interaction_counts = interaction_counts[1:]

fig = px.pie(interaction_counts, names="interaction", values="count", title="Count of protests per interaction code")
fig.show()


blm_data = df[df['assoc_actor_1'].str.contains('BLM: Black Lives Matter', na=False)]
#Isolating all the rows with the mention of BLM in the associated actor

yearly_sub_event_counts = blm_data.groupby([blm_data['event_date'].dt.year, 'sub_event_type']).size().unstack(fill_value=0)

sns.heatmap(yearly_sub_event_counts, cmap='viridis', annot=True, fmt='d')
plt.title('Yearly Count of "BLM: Black Lives Matter" by Sub-event Type')
plt.xlabel('Sub-event Type')
plt.ylabel('Year')
plt.show()


blm_data.columns

Index(['data_id', 'iso', 'event_id_cnty', 'event_id_no_cnty', 'event_date',
       'year', 'time_precision', 'event_type', 'sub_event_type', 'actor1',
       'assoc_actor_1', 'inter1', 'actor2', 'assoc_actor_2', 'inter2',
       'interaction', 'region', 'country', 'admin1', 'admin2', 'location',
       'latitude', 'longitude', 'geo_precision', 'source', 'source_scale',
       'notes', 'fatalities', 'timestamp', 'iso3', 'year_month'],
      dtype='object')


blm_data['year_month'] = blm_data['event_date'].dt.strftime('%Y-%m')
#creating a year-month for ease of analysis from the converted event_date

monthly_protest_counts = blm_data.groupby(['year_month','admin1']).size().reset_index(name='protest_count')
#per month protest counts coloured by admin1 to check for a possible correlation

fig = px.bar(monthly_protest_counts, x='admin1', y='protest_count', color = 'year_month',
             title='BLM State Wise Protest Count Per Year_Month',
             labels={'year_month': 'Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\3871097022.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


def analyze_sentiment(text):
    return TextBlob(text).sentiment.polarity
    
blm_data = df[df['assoc_actor_1'].str.contains('BLM: Black Lives Matter', na=False) & df['notes'].notna()]

blm_data['sentiment'] = blm_data['notes'].apply(analyze_sentiment)

wordcloud_text = ' '.join(blm_data['notes'])

wordcloud = WordCloud(width=800, height=400, background_color='white').generate(wordcloud_text)
plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud for "BLM: Black Lives Matter" in Assoc_actor_1')
plt.show()

plt.figure(figsize=(8, 5))
plt.hist(blm_data['sentiment'], bins=20, color='skyblue')
plt.title('Sentiment Distribution for "BLM: Black Lives Matter" in Assoc_actor_1')
plt.xlabel('Sentiment Polarity')
plt.ylabel('Frequency')
plt.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\2395244002.py:6: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


blm_source_counts = blm_data['source'].value_counts() #Getting the counts of the sources
blm_source_10 = blm_source_counts[:10] #As there are many sources, looking here at the top 10

plt.figure(figsize=(10, 5))
blm_source_10.plot(kind='bar', color='skyblue')
plt.title('Count of "BLM: Black Lives Matter" by Source Scale')
plt.xlabel('Source Scale')
plt.ylabel('Count')
plt.xticks(rotation=90)
plt.show()


fig = px.scatter_mapbox(blm_data, lat="latitude", lon="longitude", hover_name="location", zoom=3, height=500)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(title='Map of "BLM: Black Lives Matter" Instances', margin={"r":0,"t":30,"l":0,"b":0})
fig.show()


students_data = df[df['assoc_actor_1'].str.contains('Students', na=False)]

yearly_sub_event_counts = students_data.groupby([students_data['event_date'].dt.year, 'sub_event_type']).size().unstack(fill_value=0)

plt.figure(figsize=(12, 6))
sns.heatmap(yearly_sub_event_counts, cmap='viridis', annot=True, fmt='d')
plt.title('Yearly Count of "Students (United States)" by Sub-event Type')
plt.xlabel('Sub-event Type')
plt.ylabel('Year')
plt.show()


students_data.columns

Index(['data_id', 'iso', 'event_id_cnty', 'event_id_no_cnty', 'event_date',
       'year', 'time_precision', 'event_type', 'sub_event_type', 'actor1',
       'assoc_actor_1', 'inter1', 'actor2', 'assoc_actor_2', 'inter2',
       'interaction', 'region', 'country', 'admin1', 'admin2', 'location',
       'latitude', 'longitude', 'geo_precision', 'source', 'source_scale',
       'notes', 'fatalities', 'timestamp', 'iso3', 'year_month'],
      dtype='object')


students_data['month'] = students_data['event_date'].dt.strftime('%m')

monthly_protest_counts = students_data.groupby(['month','admin1']).size().reset_index(name='protest_count')

fig = px.bar(monthly_protest_counts, x='admin1', y='protest_count', color = 'month',
             title='students Monthly Protest Count Per Year',
             labels={'year_month': 'Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\3831895891.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


students_notes = ' '.join(df[df['assoc_actor_1'].str.contains('Students', na=False)]['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color='white').generate(students_notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud for "Students" in Assoc_actor_1')
plt.show()


students_source_counts = students_data['source'].value_counts()
students_source_10 = students_source_counts[:10]

plt.figure(figsize=(10, 5))
students_source_10.plot(kind='bar', color='skyblue')
plt.title('Count of "students" by Source ')
plt.xlabel('Source')
plt.ylabel('Count')
plt.xticks(rotation=90)
plt.show()


fig = px.scatter_mapbox(students_data, lat="latitude", lon="longitude", hover_name="location", zoom=3, height=500)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(title='Map of "students" Instances', margin={"r":0,"t":30,"l":0,"b":0})
fig.show()


labour_group_data = df[df['assoc_actor_1'].str.contains('Labour Group', na=False)]

yearly_sub_event_counts = labour_group_data.groupby([labour_group_data['event_date'].dt.year, 'sub_event_type']).size().unstack(fill_value=0)

plt.figure(figsize=(12, 6))
sns.heatmap(yearly_sub_event_counts, cmap='viridis', annot=True, fmt='d')
plt.title('Yearly Count of "labour_group (United States)" by Sub-event Type')
plt.xlabel('Sub-event Type')
plt.ylabel('Year')
plt.show()


labour_group_data['month'] = labour_group_data['event_date'].dt.strftime('%m')

monthly_protest_counts = labour_group_data.groupby(['month','admin1']).size().reset_index(name='protest_count')

fig = px.bar(monthly_protest_counts, x='admin1', y='protest_count', color = 'month',
             title='labour_group Monthly Protest Count Per Year',
             labels={'year_month': 'Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\3227680976.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


labour_group_notes = ' '.join(df[df['assoc_actor_1'].str.contains('Labour Group', na=False)]['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color='white').generate(labour_group_notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud for "labour_group" in Assoc_actor_1')
plt.show()


labour_group_source_counts = labour_group_data['source'].value_counts()
labour_group_source_10 = labour_group_source_counts[:10]

plt.figure(figsize=(10, 5))
labour_group_source_10.plot(kind='bar', color='skyblue')
plt.title('Count of "labour_group" by Source Scale')
plt.xlabel('Source Scale')
plt.ylabel('Count')
plt.xticks(rotation=90)
plt.show()


fig = px.scatter_mapbox(labour_group_data, lat="latitude", lon="longitude", hover_name="location", zoom=3, height=500)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(title='Map of "Labour Group" Instances', margin={"r":0,"t":30,"l":0,"b":0})
fig.show()


women_data = df[df['assoc_actor_1'].str.contains('Women', na=False)]

yearly_sub_event_counts = women_data.groupby([women_data['event_date'].dt.year, 'sub_event_type']).size().unstack(fill_value=0)

plt.figure(figsize=(12, 6))
sns.heatmap(yearly_sub_event_counts, cmap='viridis', annot=True, fmt='d')
plt.title('Yearly Count of "Women (United States)" by Sub-event Type')
plt.xlabel('Sub-event Type')
plt.ylabel('Year')
plt.show()


women_data['month'] = women_data['event_date'].dt.strftime('%m')

monthly_protest_counts = women_data.groupby(['month','admin1']).size().reset_index(name='protest_count')

fig = px.bar(monthly_protest_counts, x='admin1', y='protest_count', color = 'month',
             title='women Monthly Protest Count Per Year',
             labels={'year_month': 'Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\3970196015.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


women_notes = ' '.join(df[df['assoc_actor_1'].str.contains('Women', na=False)]['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color='white').generate(women_notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud for "women" in Assoc_actor_1')
plt.show()


women_source_counts = women_data['source'].value_counts()
women_source_count_10 = women_source_counts[:10]

plt.figure(figsize=(10, 5))
women_source_count_10.plot(kind='bar', color='skyblue')
plt.title('Count of "women" by Source Scale')
plt.xlabel('Source Scale')
plt.ylabel('Count')
plt.xticks(rotation=90)
plt.show()


fig = px.scatter_mapbox(women_data, lat="latitude", lon="longitude", hover_name="location", zoom=3, height=500)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(title='Map of "Woman" Instances', margin={"r":0,"t":30,"l":0,"b":0})
fig.show()


govt_data = df[df['assoc_actor_1'].str.contains('Government of the United States', na=False)]

yearly_sub_event_counts = govt_data.groupby([govt_data['event_date'].dt.year, 'sub_event_type']).size().unstack(fill_value=0)

plt.figure(figsize=(12, 6))
sns.heatmap(yearly_sub_event_counts, cmap='viridis', annot=True, fmt='d')
plt.title('Yearly Count of "Govt (United States)" by Sub-event Type')
plt.xlabel('Sub-event Type')
plt.ylabel('Year')
plt.show()


govt_data['month'] = govt_data['event_date'].dt.strftime('%m')

monthly_protest_counts = govt_data.groupby(['month','admin1']).size().reset_index(name='protest_count')

fig = px.bar(monthly_protest_counts, x='admin1', y='protest_count', color = 'month',
             title='govt Monthly Protest Count Per Year',
             labels={'year_month': 'Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\2211750188.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


govt_notes = ' '.join(df[df['assoc_actor_1'].str.contains('Government of the United States', na=False)]['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color='white').generate(govt_notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud for "govt" in Assoc_actor_1')
plt.show()


govt_source_counts = govt_data['source'].value_counts()
govt_source_count_10 = govt_source_counts[:10]

plt.figure(figsize=(10, 5))
govt_source_count_10.plot(kind='bar', color='skyblue')
plt.title('Count of "govt" by Source Scale')
plt.xlabel('Source Scale')
plt.ylabel('Count')
plt.xticks(rotation=90)
plt.show()


fig = px.scatter_mapbox(govt_data, lat="latitude", lon="longitude", hover_name="location", zoom=3, height=500)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(title='Map of "Government of the United States" Instances', margin={"r":0,"t":30,"l":0,"b":0})
fig.show()


blm_data['assoc_actor'] = 'BLM'
students_data['assoc_actor'] = 'Students'
labour_group_data['assoc_actor'] = 'Labour Group'
women_data['assoc_actor'] = 'Women'
govt_data['assoc_actor'] = 'Government'

# Combining all the maps gathered from the previous data to visualize them at the same glance
# to undestand and discern the differences between the top 5 associated actors

combined_data = pd.concat([blm_data, students_data, labour_group_data, women_data, govt_data])


fig = px.scatter_mapbox(combined_data, lat="latitude", lon="longitude",
                        color="assoc_actor", hover_name="location",
                        zoom=3, height=500, title='Combined Map of Various Associated Actors ')

fig.update_layout(mapbox_style="open-street-map",
                  mapbox_zoom=3,
                  mapbox_center={"lat": 37.0902, "lon": -95.7129},
                  margin={"r":0,"t":0,"l":0,"b":0},
                  legend_title_text='assoc_actor')

fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\812147166.py:1: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\812147166.py:2: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\812147166.py:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\812147166.py:4: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\812147166.py:5: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


df_filtered = df[df['event_date'].dt.year == 2020] #filtering our for the year 2020

df_filtered['year_month'] = df_filtered['event_date'].dt.strftime('%Y-%m')
#For a month-year value taking the Y/m from the datetype converted event_date

monthly_protest_counts = df_filtered.groupby(['year_month', 'event_type']).size().to_frame(name='protest_count').reset_index()
#grouping by year_month to get the protest counts per month by the event_type color

fig = px.bar(monthly_protest_counts, x='year_month', y='protest_count', color = 'event_type',
             title='Monthly Protest Count in 2020',
             labels={'year_month': 'Year-Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\1430075707.py:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


df_filtered_riot = df_filtered[df_filtered['event_type'] == 'Riots']
df_filtered_protests = df_filtered[df_filtered['event_type'] == 'Protests']


df_filtered_riot_day = df_filtered_riot.groupby('event_date')['event_type'].count().reset_index(name='riot_count')
fig = px.bar(df_filtered_riot_day, x='event_date', y='riot_count', title='Daily Riot Count in 2020')
#Daily Riot count for the year 2020
fig.show()


df_filtered_protests_day = df_filtered_protests.groupby('event_date')['event_type'].count().reset_index(name='protest_count')
fig = px.bar(df_filtered_protests_day, x='event_date', y='protest_count', title='Daily Protest Count in 2020')
#Daily protest count for the year 2020
fig.show()


df = pd.read_csv("North_America-United_States.csv")
df["event_date"] = pd.to_datetime(df["event_date"])

filtered_df = df[(df['year'] == 2020) &
                        (df['event_type'] == 'Riots')]
#Selecting a subset of only the Riots in 2020

fig = make_subplots(rows=2, cols=2, specs=[[{'type': 'domain'}, {'type': 'domain'}],
                                           [{'type': 'domain'}, {'type': 'domain'}]],
                    subplot_titles=('May 2020', 'June 2020', 'July 2020', 'August 2020'))

months = ['May', 'June', 'July', 'August']
subplot_positions = [(1, 1), (1, 2), (2, 1), (2, 2)]

for month, position in zip(months, subplot_positions):

    month_df = filtered_df[filtered_df['event_date'].dt.strftime('%B') == month]

    assoc_actor_counts = month_df['assoc_actor_1'].str.split(';').explode().value_counts()
#the associated actors in many rows are seperated by ';'. Dealing with that to get the accurate value counts
#of the associated actors.

    top_contributors = assoc_actor_counts.head(5)

    labels = top_contributors.index
    values = top_contributors.values

    fig.add_trace(go.Pie(labels=labels, values=values, name=month), row=position[0], col=position[1])

fig.update_layout(
    title=dict(text='Top 5 Contributors in Associated Actors for Riots (May-Aug 2020)', x=0.5),
    font=dict(size=12),
    width=800,
    height=600
)

fig.show()


filtered_df = df[(df['year'] == 2020) &
                        (df['event_type'] == 'Protests')]
#Selecting a subset of only the Protests in 2020

fig = make_subplots(rows=2, cols=2, specs=[[{'type': 'domain'}, {'type': 'domain'}],
                                           [{'type': 'domain'}, {'type': 'domain'}]],
                    subplot_titles=('May 2020', 'June 2020', 'July 2020', 'August 2020'))
#Creating subplot for easy visibility

months = ['May', 'June', 'July', 'August']
subplot_positions = [(1, 1), (1, 2), (2, 1), (2, 2)]

for month, position in zip(months, subplot_positions):
    month_df = filtered_df[filtered_df['event_date'].dt.strftime('%B') == month]

    assoc_actor_counts = month_df['assoc_actor_1'].str.split(';').explode().value_counts()
#the associated actors in many rows are seperated by ';'. Dealing with that to get the accurate value counts
#of the associated actors.

    top_contributors = assoc_actor_counts.head(5)

    labels = top_contributors.index
    values = top_contributors.values

    fig.add_trace(go.Pie(labels=labels, values=values, name=month), row=position[0], col=position[1])

fig.update_layout(
    title=dict(text='Top 5 Contributors in Associated Actors for Protests (May-Aug 2020)', x=0.5),
    font=dict(size=12),
    width=800,
    height=600
)

fig.show()


filtered_df = df[(df['event_date'].dt.year == 2020) &
                 (df['event_date'].dt.month.isin([5, 6, 7, 8]))]
#Choosing the months when there was a peak in activity in 2020

notes = ' '.join(filtered_df['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color ='white').generate(notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud of Notes (May-August 2020)')
plt.show()


df_filtered = df[df['event_date'].dt.year == 2021]

df_filtered['year_month'] = df_filtered['event_date'].dt.strftime('%Y-%m')

monthly_protest_counts = df_filtered.groupby(['year_month', 'event_type']).size().to_frame(name='protest_count').reset_index()
#Like we previously did, getting the event_count for each month in 2021 coloured by event_type

fig = px.bar(monthly_protest_counts, x='year_month', y='protest_count', color = 'event_type',
             title='Monthly Protest Count in 2021',
             labels={'year_month': 'Year-Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\2835233933.py:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


df_filtered_riot = df_filtered[df_filtered['event_type'] == 'Riots']
df_filtered_protests = df_filtered[df_filtered['event_type'] == 'Protests']


df_filtered_riot_day = df_filtered_riot.groupby('event_date')['event_type'].count().reset_index(name='riot_count')
fig = px.bar(df_filtered_riot_day, x='event_date', y='riot_count', title='Daily Riot Count in 2021')
#Daily Riot count in the year 2021
fig.show()


df_filtered_protests_day = df_filtered_protests.groupby('event_date')['event_type'].count().reset_index(name='protest_count')
fig = px.bar(df_filtered_protests_day, x='event_date', y='protest_count', title='Daily Protest Count in 2021')
# Daily protest count in the year 2021
fig.show()


filtered_df = df[(df['year'] == 2021) &
                        (df['event_type'] == 'Protests')]

fig = make_subplots(rows=2, cols=2, specs=[[{'type': 'domain'}, {'type': 'domain'}],
                                           [{'type': 'domain'}, {'type': 'domain'}]],
                    subplot_titles=('January 2021', 'March 2021', 'May 2021', 'October 2021'))

months = ['January', 'March', 'May', 'October']
subplot_positions = [(1, 1), (1, 2), (2, 1), (2, 2)]

for month, position in zip(months, subplot_positions):
    month_df = filtered_df[filtered_df['event_date'].dt.strftime('%B') == month]

    assoc_actor_counts = month_df['assoc_actor_1'].str.split(';').explode().value_counts()
    #similarly as above instances splitting and getting the counts for the associated actors

    top_contributors = assoc_actor_counts.head(5)

    labels = top_contributors.index
    values = top_contributors.values

    fig.add_trace(go.Pie(labels=labels, values=values, name=month), row=position[0], col=position[1])

fig.update_layout(
    title=dict(text='Top 5 Contributors in Associated Actors for Protests (January, March, May, Oct 2021)', x=0.5),
    font=dict(size=12),
    width=800,
    height=600
)

fig.show()


filtered_df = df[(df['event_date'].dt.year == 2021) &
                 (df['event_date'].dt.month.isin([1, 3, 5, 10]))]#selecting the months which showed peak in activity

notes = ' '.join(filtered_df['notes'].dropna())#combining information from the notes for wordcloud

wordcloud = WordCloud(width=800, height=400, background_color ='white').generate(notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud of Notes (January, March, May, Oct 2021)')
plt.show()


df_filtered = df[df['event_date'].dt.year == 2022] # selecting only the year 2022

df_filtered['year_month'] = df_filtered['event_date'].dt.strftime('%Y-%m') #Creating a year-month for understanding

monthly_protest_counts = df_filtered.groupby(['year_month', 'event_type']).size().to_frame(name='protest_count').reset_index()
#getting the count of protests per month in the year 2022

fig = px.bar(monthly_protest_counts, x='year_month', y='protest_count', color = 'event_type',
             title='Monthly Protest Count in 2022',
             labels={'year_month': 'Year-Month', 'protest_count': 'Number of Protests'})
fig.update_xaxes(tickangle=-45)
fig.show()

C:\Users\ashwi\AppData\Local\Temp\ipykernel_9088\2695222818.py:3: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


df_filtered_riot = df_filtered[df_filtered['event_type'] == 'Riots']
df_filtered_protests = df_filtered[df_filtered['event_type'] == 'Protests']


sub_event_counts = df_filtered_protests['sub_event_type'].value_counts().reset_index()
sub_event_counts.columns = ['sub_event_type', 'count'] # Getting the Sub_event_types and their counts.
fig = px.pie(sub_event_counts, names="sub_event_type", values="count",hole = 0.3, title="Count of Protests Sub Event types")
fig.show()

# contribution of sub_event type for the year 2022.


df_filtered_riot_day = df_filtered_riot.groupby('event_date')['event_type'].count().reset_index(name='riot_count')
fig = px.bar(df_filtered_riot_day, x='event_date', y='riot_count', title='Daily Riot Count in 2022')
#Daily riot count in 2022 to see and analyze , time of the year for the spikes
fig.show()


df_filtered_protests_day = df_filtered_protests.groupby('event_date')['event_type'].count().reset_index(name='protest_count')
fig = px.bar(df_filtered_protests_day, x='event_date', y='protest_count', title='Daily Protest Count in 2022')
#Daily Protest count in 2022 to see and analyze , time of the year for the spikes
fig.show()


filtered_df = df[(df['year'] == 2022) &
                        (df['event_type'] == 'Protests')]

# we are filtering out all the protest instances in 2022, to understand the Associated actors at play.

fig = make_subplots(rows=2, cols=2, specs=[[{'type': 'domain'}, {'type': 'domain'}],
                                           [{'type': 'domain'}, {'type': 'domain'}]],
                    subplot_titles=('January 2022', 'March 2022', 'May 2022', 'June 2022'))

months = ['January', 'March', 'May', 'June']
subplot_positions = [(1, 1), (1, 2), (2, 1), (2, 2)]

for month, position in zip(months, subplot_positions):

    month_df = filtered_df[filtered_df['event_date'].dt.strftime('%B') == month]

    assoc_actor_counts = month_df['assoc_actor_1'].str.split(';').explode().value_counts()

    top_contributors = assoc_actor_counts.head(5)

# Through this for loop we are considering 1 month at a time, dealing with the associated actors appropriately
# Using that we are getting the Associated actor counts for the top 4 months

    labels = top_contributors.index
    values = top_contributors.values

    fig.add_trace(go.Pie(labels=labels, values=values, name=month), row=position[0], col=position[1])

fig.update_layout(
    title=dict(text='Top 5 Contributors in Associated Actors for Protests (January, March, May, Jun 2022)', x=0.5),
    font=dict(size=12),
    width=800,
    height=600
)

fig.show()


filtered_df = df[(df['event_date'].dt.year == 2022) &
                 (df['event_date'].dt.month.isin([1, 3, 5, 6]))]

# Like we have done previously, we are generating a word clound for the top 4 peak months in the year 2022

notes = ' '.join(filtered_df['notes'].dropna())

wordcloud = WordCloud(width=800, height=400, background_color ='white').generate(notes)

plt.figure(figsize=(10, 5))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis('off')
plt.title('Word Cloud of Notes (January, March, May, Jun 2022)')
plt.show()

	data_id	iso	event_id_cnty	event_id_no_cnty	event_date	year	time_precision	event_type	sub_event_type	actor1	...	location	latitude	longitude	geo_precision	source	source_scale	notes	timestamp	iso3
0	9383068	840	USA44621	44621	24 June 2022	2022	1	Protests	Excessive force against protesters	Protesters (United States)	...	Cedar Rapids	41.9738	-91.6768	1	KCRG-TV9; Network Contagion Research Institute...	Local partner-National	On 24 June 2022, 'pro-choice' protesters gathe...	1656458658	USA
1	9383099	840	USA44775	44775	24 June 2022	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Fayetteville	36.0626	-94.1574	1	Southwest Times Record; 40/29	Subnational	On 24 June 2022, well over 100 people includin...	1656458658	USA
2	9383100	840	USA44783	44783	24 June 2022	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Eureka	40.8040	-124.1654	1	Lost Coast Outpost; North Coast Journal	Subnational	On 24 June 2022, hundreds of people rallied ou...	1656458658	USA
3	9383101	840	USA44788	44788	24 June 2022	2022	1	Protests	Peaceful protest	Protesters (United States)	...	West Hollywood	34.0900	-118.3617	1	CBS News; Liveuamap	Local partner-New media	On 24 June 2022, hundreds of people rallied in...	1656458658	USA
4	9383102	840	USA44793	44793	24 June 2022	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Monterey	36.6031	-121.8936	1	Monterey County Herald; KSBW8	Subnational	On 24 June 2022, hundreds of people rallied at...	1656458658	USA

	data_id	iso	event_id_cnty	event_id_no_cnty	event_date	year	time_precision	event_type	sub_event_type	actor1	...	location	latitude	longitude	geo_precision	source	source_scale	notes	timestamp	iso3
0	9383068	840	USA44621	44621	2022-06-24	2022	1	Protests	Excessive force against protesters	Protesters (United States)	...	Cedar Rapids	41.9738	-91.6768	1	KCRG-TV9; Network Contagion Research Institute...	Local partner-National	On 24 June 2022, 'pro-choice' protesters gathe...	1656458658	USA
1	9383099	840	USA44775	44775	2022-06-24	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Fayetteville	36.0626	-94.1574	1	Southwest Times Record; 40/29	Subnational	On 24 June 2022, well over 100 people includin...	1656458658	USA
2	9383100	840	USA44783	44783	2022-06-24	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Eureka	40.8040	-124.1654	1	Lost Coast Outpost; North Coast Journal	Subnational	On 24 June 2022, hundreds of people rallied ou...	1656458658	USA
3	9383101	840	USA44788	44788	2022-06-24	2022	1	Protests	Peaceful protest	Protesters (United States)	...	West Hollywood	34.0900	-118.3617	1	CBS News; Liveuamap	Local partner-New media	On 24 June 2022, hundreds of people rallied in...	1656458658	USA
4	9383102	840	USA44793	44793	2022-06-24	2022	1	Protests	Peaceful protest	Protesters (United States)	...	Monterey	36.6031	-121.8936	1	Monterey County Herald; KSBW8	Subnational	On 24 June 2022, hundreds of people rallied at...	1656458658	USA

	data_id	iso	event_id_no_cnty	event_date	time_precision	inter1	inter2	interaction	latitude	longitude	geo_precision	fatalities	timestamp
count	4.223800e+04	42238.0	42238.000000	42238	42238.000000	42238.000000	42238.000000	42238.000000	42238.000000	42238.000000	42238.000000	42238.000000	4.223800e+04
mean	8.266746e+06	840.0	22430.918959	2021-02-01 18:58:12.779014144	1.005422	5.974620	0.330177	58.229627	38.370137	-92.543588	1.007955	0.000710	1.627587e+09
min	7.593655e+06	840.0	1.000000	2020-01-01 00:00:00	1.000000	5.000000	0.000000	15.000000	19.442700	-166.534100	1.000000	0.000000	1.612546e+09
25%	7.616716e+06	840.0	11024.250000	2020-06-20 00:00:00	1.000000	6.000000	0.000000	60.000000	34.799800	-106.651100	1.000000	0.000000	1.612547e+09
50%	8.265290e+06	840.0	22444.500000	2020-11-17 00:00:00	1.000000	6.000000	0.000000	60.000000	39.219100	-86.443600	1.000000	0.000000	1.624483e+09
75%	8.680164e+06	840.0	33741.750000	2021-08-18 00:00:00	1.000000	6.000000	0.000000	60.000000	41.850000	-77.615500	1.000000	0.000000	1.638920e+09
max	9.383708e+06	840.0	45052.000000	2022-06-24 00:00:00	3.000000	6.000000	8.000000	68.000000	71.290600	-67.461400	3.000000	3.000000	1.656459e+09
std	5.543203e+05	0.0	13078.246226	NaN	0.074075	0.157278	1.280650	9.161994	5.195800	18.214649	0.097962	0.030766	1.388826e+07

	Actor Pair	Frequency
0	(BLM: Black Lives Matter, Pro-Police Group (Un...	173
1	(Pro-Police Group (United States), BLM: Black ...	105
2	(Back the Blue, BLM: Black Lives Matter)	60
3	(BLM: Black Lives Matter, Back the Blue)	53
4	(BLM: Black Lives Matter, Civilians (United St...	38
5	(Proud Boys, Antifa (United States))	36
6	(Students (United States), Students (United St...	35
7	(Women (United States), Students (United States))	31
8	(Women (United States), Christian Group (Unite...	30
9	(BLM: Black Lives Matter, Journalists (United ...	30

Table of Contents¶

Data Overview¶

Pre-processing¶

Univariate Analysis¶

Data Cleaning¶

Associate Actor and top 5 associate Actors Analysis¶

Relation between Actor 1 and Actor 2¶

BLM¶

Students¶

Labour Group¶

Women¶

Government of United States¶

Map¶

Year on Year Analysis¶

2020¶

2021¶

2022¶

Per Section Analysis¶

Pre-processing and EDA¶

BLM¶

Heatmap of Yearly Count by Sub-Event Type¶

Bar Chart of Monthly Protest Count Per State¶

Word Cloud for "BLM: Black Lives Matter" in Assoc_actor_1¶

Bar Chart of Protest Count by Source Scale¶

Map of "Black Lives Matter" Instances¶

Students¶

Geographical Distribution Map¶

Heatmap of Yearly Count by Sub-Event Type¶

Monthly Protest Count Per State Bar Chart¶

Word Cloud of Associated Actors¶

Bar Chart of Protest Count by Source¶

Geographical Distribution Map¶

Labour Group¶

Heatmap of Yearly Count by Sub-Event Type¶

Bar Chart of Monthly Protest Count Per State¶

Word Cloud for Labour Group¶

Bar Chart of Protest Count by Source Scale¶

Map of "Labour Group" Instances¶

Women¶

Assoc_actor 1 and Assoc_actor_2 pair analysis¶

Govt USA¶

Heatmap of Yearly Count by Sub-Event Type¶

Bar Chart of Monthly Protest Count Per State¶

Word Cloud for "Govt" in Assoc_actor_1¶

Bar Chart of Protest Count by Source Scale¶

Map of "Government of the United States" Instances¶

Map¶

Geographic Distribution and Density¶

Intensity and Overlaps¶

Inferences and Implications¶

2020¶

Monthly Protest Count in 2020¶

Daily Riot Count in 2020¶

Daily Protest Count in 2020¶

Top 5 Contributors in Associated Actors for Riots (May-August 2020)¶

Top 5 Contributors in Associated Actors for Protests (May-August 2020)¶

Word Cloud of Notes (May-August 2020)¶

Overall Analysis¶

2021¶

2022¶